The world’s second quickest supercomputer — it was the quickest, earlier than its rival machine got here on-line earlier this month — has created probably the most advanced laptop simulation of the universe so far. The aim of this simulation is to check what researchers describe as “cosmological hydrodynamics.”
The supercomputer is called Frontier, lives at Oak Ridge National Laboratory — and is a beast of a tool. Built to be the primary exascale supercomputer, it could possibly carry out as much as 1.1 exaFLOPS, which is the same as 1.1 quintillion (10^18, or 1,100,000,000,000,000,000) floating-point operations per second. It’s constructed from 9,472 AMD central processing models (CPUs) and 37,888 AMD graphics processing models (GPUs). (That’s a staggering quantity of each CPUs and GPUs). Frontier was the quickest supercomputer on the earth till one other supercomputer, named El Capitan and situated on the Lawrence Livermore National Laboratory, overtook it with 1.742 exaFLOPS in November 2024, in accordance with the AMD web site.
A staff from the U.S. Department of Energy’s Argonne National Laboratory in Illinois, led by Argonnes division director for computational science, Salman Habib, used its Hardware/Hybrid Accelerated Cosmology Code (HACC) on Frontier. First developed about 15 years in the past, HACC fashions the evolution of the universe, and its code was written in order that it could possibly be tailored for no matter is the quickest supercomputer at any given time.
HACC was initially deployed on petascale supercomputers (able to quadrillions of FLOPS), that are much less highly effective than Frontier and El Capitan. For instance, one undertaking by the Argonne staff was to make use of HACC to mannequin three totally different cosmologies on the Summit supercomputer, a petascale laptop that was the quickest on the earth between November 2018 and June 2020. The three simulations have been named after planets from the Star Trek universe: the Qo’nos simulation (named after the Klingon homeworld) modeled the universe utilizing the usual mannequin of cosmology (which includes calculations about each darkish vitality and chilly darkish matter); the Vulcan simulation included large neutrinos, and the Ferenginar simulation (named after the Ferengi homeworld) explored a universe the place darkish vitality wasn’t fixed, however modified over time. The outcomes from the Summit simulations confirmed that when darkish vitality varies it could possibly result in stronger clustering of galaxies within the early universe — one thing astronomers might search for in high-redshift galaxy surveys, that means galaxy surveys that look into pockets of the extraordinarily distant universe.
Related: Largest-ever laptop simulation of the universe escalates cosmology dilemma
In some ways, the ensuing simulations are as giant and detailed as a serious deep astronomical survey, such because the Sloan Digital Sky Survey, or the forthcoming surveys to be performed by the Vera C. Rubin Observatory.
However, the Summit simulations have been “gravity solely” simulations, aka they weren’t highly effective sufficient to incorporate different forces or results.
“If we have been to simulate a big chunk of the universe surveyed by one of many huge telescopes such because the Rubin Observatory in Chile, you are speaking about taking a look at enormous chunks of time — billions of years of enlargement,” stated Habib in a assertion. “Until not too long ago, we couldn’t even think about doing such a big simulation like that besides within the gravity-only approximation.”
More highly effective simulations to incorporate different forces and results in addition to gravity have been wanted, and that’s the place Frontier, supported by a $1.8 billion Department of Energy undertaking referred to as ExaSky to fund exascale computing, is available in.
In the normal mannequin of cosmology, two parts dominate: darkish matter and darkish vitality. The stuff that you just and I are constructed from — so-called baryonic matter — makes up lower than 5% of the matter and vitality within the universe.
“So, if we wish to know what the universe is as much as, we have to simulate each of this stuff [dark matter and dark energy] … gravity in addition to all the opposite physics together with sizzling fuel, and the formation of stars, black holes and galaxies,” stated Habib. He calls it “the astrophysical kitchen sink, so to talk. These … are what we name cosmological hydrodynamics simulations.”
In order to arrange HACC for working on Frontier, the necessities of the ExaSky undertaking have been that HACC should run at the least 50 instances quicker on Frontier than it did on Titan, which was the world’s quickest supercomputer in 2012 when HACC was initially being developed. During its dry run on Frontier, HACC blew that requirement out of the water, proving to be almost 300 instances quicker.
Bronson Messer, who’s the director of science on the Oak Ridge Leadership Computing Facility, credit HACC for including the “bodily realism of together with the baryons and all the opposite dynamic physics that makes this simulation a real tour de drive for Frontier.”
The simulations, which shall be made accessible to the astronomical neighborhood, will permit researchers to discover their cosmological fashions, asking questions in regards to the nature of darkish matter, the energy of darkish vitality, or pursuing various fashions of gravity corresponding to Modified Newtonian Dynamics (MOND). The simulations can then be in comparison with what actual astronomical surveys are discovering, to find out which mannequin most closely fits the observations.
Which all raises one fascinating philosophical query: If computing energy continues to extend, our simulations of the universe will develop extra detailed and correct in parallel, however the place does it finish?
It has been recommended by some scientific thinkers that we ourselves would possibly dwell in a simulation, maybe because the product of any individual else’s try to mannequin actuality. And if we’re creating simulations whereas being inside a simulation, then maybe it truly is turtles all the best way down.
